Survival after breast-conserving surgery versus mastectomy in mammary Paget disease: A propensity-matched SEER population-based analysis

Mastectomy has traditionally been the dominant surgical approach for mammary Paget disease, but its survival advantage over breast-conserving surgery (BCS) remains uncertain. We compared breast cancer-specific survival (BCSS) and overall survival (OS) after mastectomy versus BCS in a population-based SEER cohort. Female patients with mammary Paget disease diagnosed between 2000 and 2022 were identified; after exclusions, 641 surgically treated patients were included, of whom 323 underwent BCS and 318 underwent mastectomy. The primary treatment-strategy effect was estimated using inverse-probability-of-treatment weighting, with radiotherapy excluded from the propensity model as a mediator on the BCS pathway. A 1:1 propensity-matched cohort served as a sensitivity analysis. Survival was assessed using Kaplan-Meier analysis, multivariable Cox regression, Aalen-Johansen competing-risk estimation, Fine-Gray modelling, and additional robustness analyses. After matching, 228 pairs were analyzed. Five-year BCSS was 95.8% after BCS and 94.1% after mastectomy; corresponding ten-year estimates were 93.4% and 92.2% (log-rank p = 0.72). Ten-year cumulative incidence of breast cancer-specific death was 6.3% versus 7.4%, respectively, after accounting for competing non-cancer mortality. OS did not differ significantly between groups (p = 0.46). Mastectomy was not associated with improved BCSS in the primary weighted analysis (hazard ratio 1.19, 95% confidence interval 0.66-2.14, p = 0.56) or the matched sensitivity analysis (hazard ratio 1.30, 95% confidence interval 0.58-2.88, p = 0.52). Findings were concordant across competing-risk, overlap-weighted, parsimonious, penalized, and age-stratified models. In this SEER-based, propensity-adjusted analysis, mastectomy was not associated with a detectable survival advantage over BCS in women with mammary Paget disease. These findings support BCS as a reasonable option in selected patients, while prospective studies incorporating margin status, receptor biology, local recurrence, and patient-reported outcomes are needed.